Introduction

The artificial intelligence debate has entered a new phase.

For much of the past two years, the dominant question was relatively straightforward: How many jobs will artificial intelligence replace?

That question is now being overshadowed by a more fundamental concern. As frontier AI systems become increasingly capable, some of the researchers building them are questioning not only their implications for employment and productivity, but the possibility that sufficiently advanced AI could eventually create risks that extend beyond the economy and into the long-term survival of humanity.

Recent departures and public warnings from researchers working at leading AI companies have brought this debate into sharper focus. The concern expressed by some researchers is that AI development may be progressing faster than society's ability to evaluate, govern and control increasingly capable systems.

The significance of this debate extends beyond the laboratories where frontier AI is being developed. If the trajectory of AI ultimately leads to systems capable of performing increasingly sophisticated cognitive work with limited human intervention, the implications for organisations and employees could be profound even if the most extreme predictions about AI never materialise.

For business leaders, the relevant question is therefore not whether artificial intelligence will make employees "extinct". There is currently no credible basis for predicting the disappearance of human employment as a whole, nor is there a reliable timeline for such an outcome.

The more immediate question is more practical:

How much of today's work can increasingly capable AI perform, how quickly will organisations adopt it, and what will happen to the structure of employment when the economics of human labour begin to change?

For Africa, this question has an additional dimension. The continent generally has lower levels of digital infrastructure, AI adoption and exposure to automation than high-income economies. That may slow the pace of disruption. It may also create an important opportunity to prepare.

But it does not make Africa immune.

In fact, the very characteristics that have historically slowed technological adoption in many African economies could eventually determine whether the continent benefits from AI or falls further behind.

From AI safety to the future of work

The recent debate among researchers at frontier AI companies is fundamentally about AI safety and control, rather than employment.

That distinction matters.

The possibility that an advanced AI system could pose an existential risk to humanity is a different proposition from the possibility that AI could automate tasks currently performed by employees.

The two questions, however, are connected by one underlying development: the increasing capability of AI systems.

Earlier generations of workplace automation were generally narrow. Software could calculate, store, retrieve and process information, but human beings remained responsible for most forms of interpretation, communication and decision-making.

Generative AI changed the boundary.

Large language models can produce text, analyse information, write software, summarise documents, generate presentations, translate languages and support research. Newer AI systems are increasingly capable of executing sequences of tasks rather than simply responding to individual instructions.

The next stage is likely to involve greater integration of AI into organisational workflows, including systems in which AI agents undertake multi-step activities under varying degrees of human supervision.

This is already beginning to change how organisations think about work.

Boston Consulting Group's 2025 research on AI at work found that 72% of surveyed employees were using AI regularly, while only a much smaller proportion reported that AI agents were broadly integrated into organisational workflows. BCG's research argues that the greatest value is unlikely to come simply from placing AI tools on top of existing processes. Organisations will need to redesign workflows around the new technology.

This distinction is crucial.

The question is not simply whether AI can perform a task.

The question is whether the organisation can redesign the work around AI.

The employee is not a task

One of the most common mistakes in discussions about AI and employment is to treat a job as if it were a single activity.

It is not.

A financial analyst does not merely analyse spreadsheets. An HR manager does not simply prepare policies. A lawyer does not only draft documents. A marketing professional does not simply write content. A consultant does not simply produce PowerPoint presentations.

Occupations are collections of tasks.

Some of those tasks may be highly automatable. Others may require judgement, context, interpersonal interaction, physical presence, accountability or specialised expertise.

This is why the most credible current research does not support the proposition that AI will simply eliminate entire categories of employment.

The International Labour Organization's 2025 global assessment found that approximately one in four workers globally are in occupations with some exposure to generative AI. Yet only 3.3% of global employment falls within the highest exposure category. The ILO's conclusion is that transformation is more likely than complete replacement for most occupations because human input remains necessary for many tasks.

The ILO has subsequently emphasised an important limitation: exposure indicators should not be interpreted as forecasts of actual job losses. Whether an exposed task is actually automated depends on factors including technology adoption, cost, infrastructure, organisational decisions, regulation and workers' ability to adapt.

This provides a more useful framework for business leaders.

The immediate future is unlikely to be a binary contest between humans and machines.

It is more likely to involve a redistribution of tasks between humans and machines.

The more consequential question: how many employees will organisations need?

There is, however, a more difficult implication.

An organisation does not need to eliminate an entire occupation for AI to reduce employment.

Suppose an organisation previously required ten employees to complete a particular set of information-processing activities. If AI allows those ten employees to produce the same output with significantly less effort, management may eventually determine that it needs fewer employees to achieve the same level of production.

The occupation survives.

The tasks survive.

The organisation simply requires fewer people to perform them.

This distinction is critical because it shifts the conversation from job extinction to labour productivity and workforce demand.

Historically, technological progress has often increased productivity while changing the composition of employment. AI could accelerate that process because it is increasingly affecting cognitive and knowledge-based tasks that were previously difficult to automate.

The potential consequence is not necessarily mass unemployment.

It could instead be a gradual change in the ratio between human labour and technology required to produce a given level of output.

For business leaders, this is a workforce planning issue.

For HR leaders, it is an organisational design issue.

For employees, it is a skills issue.

And for governments, it is a labour-market policy issue.

What happens when AI moves up the organisational hierarchy?

The implications become more significant as AI moves beyond routine execution.

Consider the traditional organisational pyramid.

At the base are employees performing administrative and operational activities. Above them are professionals who analyse information and solve problems. Further up are managers who coordinate people and resources. At the top are executives who make strategic decisions.

Historically, technology has primarily automated activities lower in the pyramid.

AI is beginning to challenge that assumption.

It can assist with activities traditionally associated with professional and managerial work, including analysis, research, drafting, forecasting, coding, planning and decision support.

BCG's recent analysis of AI and workforce strategy identifies a shift towards AI taking over more execution while human employees increasingly focus on strategy, design, oversight and integration. It also points to a possible restructuring of traditional organisational pyramids as AI-augmented teams become more common.

This raises a particularly important issue for employers.

If AI reduces the amount of routine work required, organisations may need fewer employees at certain levels of the traditional hierarchy.

That could affect the junior talent pipeline.

For decades, entry-level employees have learned by performing relatively simple tasks before progressing to more complex responsibilities.

But if AI increasingly performs the simple tasks, organisations may need to rethink how junior employees acquire experience.

This may become one of the most important workforce challenges of the AI era.

The issue may not be that young people have no jobs.

The issue may be that some of the traditional entry points into professional careers become narrower.

The African question: does lower technology adoption provide protection?

Africa enters this discussion from a different starting point.

The continent has lower levels of digital infrastructure and technology adoption than many advanced economies. The ILO's global assessment shows a strong relationship between national income and exposure to generative AI. Approximately 34% of employment in high-income countries falls within the four exposure categories identified by the ILO, compared with approximately 11% in low-income countries.

At first glance, this appears favourable for African workers.

If AI exposure is lower, then disruption should also be lower.

There is truth in this, but it would be dangerous to interpret the finding as evidence that Africa is insulated from AI.

Lower exposure reflects, in part, the structure of African economies and the constraints on technology adoption.

Many African economies have large informal sectors, lower levels of digitalisation, weaker connectivity and greater concentrations of employment in occupations that are currently less exposed to generative AI.

These factors reduce immediate exposure.

They also reduce the productivity gains that AI can deliver.

This creates a paradox.

The same conditions that protect some African workers from AI disruption can also prevent those workers and organisations from benefiting from AI.

Africa may therefore face two risks simultaneously.

The first is eventual disruption as AI adoption increases.

The second is falling behind as other economies use AI to increase productivity.

The second risk may prove just as consequential as the first.

Africa is not necessarily behind the AI curve

It is important not to reduce Africa's technology story to the language of "lagging".

Africa is not a single technology market.

Adoption differs significantly by country, sector, organisation and demographic group.

Financial services, telecommunications and technology companies in several African markets are already experimenting with AI at a level that would challenge the idea that the continent is simply waiting for technology to arrive.

McKinsey's analysis of generative AI in Africa estimates that widespread deployment could generate between $61 billion and $103 billion in additional economic value annually across the continent, although the realisation of that value depends on adoption, infrastructure, skills and implementation.

BCG similarly argues that African countries have an opportunity to accelerate AI adoption rather than simply follow the technological path of advanced economies. Its analysis identifies talent development, AI literacy, research capacity and digital infrastructure as key priorities.

This suggests that Africa should not be viewed simply as a late recipient of AI.

It could also become an important adopter of AI-enabled business models.

The continent's history with mobile technology demonstrates that technological development does not always follow a linear path.

In some areas, African markets have bypassed legacy infrastructure and moved directly towards newer technologies.

AI could create another opportunity for this type of leapfrogging.

The diffusion question: will Africa simply adopt AI later?

Everett Rogers' Diffusion of Innovations framework provides a useful lens for understanding this issue.

Technologies tend to spread through populations in stages. Innovators and early adopters experiment first. Adoption then expands through the early majority and late majority before reaching groups that adopt more slowly.

This framework is particularly relevant to AI because adoption is not determined solely by technological capability.

It also depends on perceived value, cost, compatibility with existing systems, complexity, infrastructure and institutional readiness.

Africa's lower adoption today therefore does not necessarily indicate that African organisations will remain behind indefinitely.

Technology adoption can accelerate once costs fall, infrastructure improves and the benefits become more visible.

The mobile money experience provides a useful precedent. Countries that did not possess mature traditional banking infrastructure were able to develop large-scale digital financial ecosystems without first reproducing every stage of the banking infrastructure of advanced economies.

AI could follow a similar path.

The question is whether African organisations will be ready when the adoption curve steepens.

Ghana: a useful case study

Ghana illustrates the tension between technological opportunity and organisational readiness.

The country has made significant progress in digital financial services and technology adoption, while AI adoption is increasingly entering the corporate conversation.

The development of a national AI strategy also signals that AI is no longer being treated purely as a private-sector technology issue. It is increasingly part of the country's broader economic and development agenda.

At the organisational level, however, adoption is uneven.

Some businesses are already incorporating generative AI into research, content development, customer service, analytics and software-related activities. Others remain dependent on manual processes and have yet to develop formal AI strategies.

This creates a significant opportunity for Ghanaian employers.

The country does not need to wait for the most advanced economies to determine what the future of work looks like.

It can learn from them.

The priority should be to identify which parts of the global AI transformation are relevant to Ghanaian industries and begin preparing employees before disruption becomes unavoidable.

The timeline problem

Can we say when AI will materially change African employment?

Not with precision.

There is no credible basis for assigning a single year in which African employees will become redundant because of AI.

There is, however, enough evidence to distinguish between different time horizons.

The near term: the next one to three years

The most likely effect is task transformation.

Organisations will increasingly use AI for research, drafting, customer support, administration, analysis, recruitment and other information-intensive activities.

The immediate workforce question will be productivity.

Can five employees supported by AI produce what previously required eight?

Can a recruiter screen more candidates?

Can an analyst process more information?

Can a consultant conduct more research?

Can an HR team automate administrative processes and spend more time on strategic work?

At this stage, the strongest competitive advantage may belong to organisations that redesign work effectively rather than organisations that simply purchase AI tools.

The medium term: approximately three to seven years

The effects could become more structural.

AI agents may undertake increasingly complex sequences of work. Organisations may redesign roles around human oversight and machine execution.

Some entry-level and coordination-heavy positions could become more difficult to justify in their current form.

At the same time, demand could increase for people who can integrate AI into business processes, manage AI-enabled workflows, interpret outputs and provide human judgement.

The organisational pyramid may begin to change.

The longer term: beyond seven years

Uncertainty increases substantially.

If AI capabilities continue advancing rapidly, the boundary between tasks suitable for humans and tasks suitable for machines could shift significantly.

At that point, conventional workforce planning models may become less reliable.

However, it would be inappropriate to translate today's AI capability trends into a precise forecast of mass human unemployment.

The longer the forecasting horizon, the greater the uncertainty.

What this means for African employers

The appropriate response for African organisations is not to freeze hiring because AI may eventually replace some work.

Nor is it to ignore AI because current exposure remains relatively low.

A more strategic approach is to begin redesigning the workforce now.

Four questions should sit at the centre of that process.

1. Which tasks are most exposed?

Organisations should map jobs by task rather than treating entire occupations as either "safe" or "at risk".

This creates a more useful picture of where automation could occur.

2. Where can AI increase productivity?

Not every use of AI should be framed as a headcount reduction opportunity.

An organisation may choose to use productivity gains to increase output, enter new markets, improve customer service or allow employees to focus on higher-value activities.

3. Which capabilities will become more valuable?

As routine execution becomes easier to automate, organisations are likely to place greater value on capabilities such as problem-solving, judgement, leadership, communication, relationship management, creativity, domain expertise and AI fluency.

4. How should roles be redesigned?

The most important question is not:

"Which employees can we replace?"

It is:

"What should this role look like when AI performs the tasks that AI is better equipped to perform?"

That is a fundamentally different approach to workforce transformation.

What this means for HR

The AI transition is often presented as a technology agenda.

It is also an HR agenda.

Human resources functions will have to reconsider how organisations recruit, develop, evaluate and retain talent.

Recruitment criteria may change.

Job descriptions may change.

Performance expectations may change.

Training programmes may change.

Career paths may change.

Organisational structures may change.

Even the definition of an entry-level role may change.

This means HR cannot remain a downstream function that responds after technology decisions have already been made.

HR needs to be part of the AI strategy from the beginning.

The CHRO, CIO and business leadership will increasingly need to work together to answer one fundamental question:

What workforce does the organisation need when technology changes the way work is performed?

BCG's research similarly points to the need for organisations to combine technology deployment with workforce readiness, reskilling, communication and organisational redesign.

What this means for employees

For employees, the implications are equally significant.

The safest response is not to attempt to compete with AI at everything.

It is to understand where human capability continues to create value and learn how to use AI to extend that capability.

An employee who can use AI to research a problem, evaluate the resulting information, apply professional judgement and communicate a recommendation to a client is likely to create more value than an employee who simply performs the underlying research manually.

The distinction is therefore moving from task execution towards problem ownership.

Employees will increasingly need to demonstrate three layers of capability:

Domain expertise: understanding the work and the business.

AI fluency: understanding how to use AI effectively and responsibly.

Human judgement: knowing what to accept, what to challenge and when a human decision is required.

The combination is more valuable than any one of the three in isolation.

Is Africa in a safe zone?

No.

But neither is Africa necessarily facing the same immediate level of disruption as high-income economies.

The evidence suggests that African employment currently has lower exposure to generative AI. That gives the continent more time.

The strategic question is what Africa does with that time.

If lower exposure becomes an excuse for inaction, the continent could eventually face a double disadvantage: fewer AI-driven productivity gains today and greater difficulty adapting when adoption accelerates.

If the same period is used to strengthen digital infrastructure, develop AI capability, redesign education and training, improve organisational readiness and build locally relevant AI applications, Africa could approach the next phase of AI from a position of greater preparedness.

The difference will not be determined by whether Africa adopts AI first or last.

It will be determined by how prepared it is when adoption becomes economically unavoidable.

The strategic implication

The most important conclusion from the current AI debate is not that employees are becoming obsolete.

It is that the economic relationship between people, technology and productivity is being renegotiated.

The recent warnings from researchers concerned about the long-term risks of increasingly capable AI systems have expanded the conversation about what AI could ultimately mean for humanity.

For business leaders, however, the immediate issue is closer to home.

AI does not need to reach a hypothetical level of superintelligence before it changes the workforce.

It only needs to become sufficiently capable, sufficiently affordable and sufficiently reliable for organisations to redesign how work is performed.

That process has already begun.

For Africa, the current lower level of exposure should be understood as a window for preparation, not a guarantee of protection.

The organisations that benefit most from AI will not necessarily be those that automate the fastest. They will be those that understand where technology creates value, where human capability remains essential, and how the two can be deliberately combined.

For employees, the implication is equally clear. The future of work is unlikely to belong exclusively to people who can compete with machines, nor exclusively to machines that can replace people.

It will increasingly belong to people and organisations that know how to work with machines while retaining the judgement, accountability, creativity and human connection that technology cannot simply be assumed to provide.

The question facing African organisations is therefore not whether AI will eventually arrive.

It is already arriving.

The more consequential question is whether our organisations, our HR systems and our workforce will be ready when the pace of adoption accelerates.

The future of work will not be determined by AI alone. It will be determined by the choices organisations make about people, technology and work.

DayStart Executive Solutions

At DayStart Executive Solutions, we believe that preparing for the future of work requires both sides of the equation: preparing executives for organisations and preparing organisations for executives.

AI makes that mission more relevant, not less.

The organisations that will navigate the transition successfully will need to rethink workforce planning, recruitment, employee development, performance management, organisational design and leadership.

And the professionals who will thrive will need more than technical competence. They will need the capacity to learn, adapt, work with technology and create value in an environment where the definition of work itself is changing.